Higher-dimensional gcd matrices (Q1190114)

The author evaluates the higher-dimensional determinants of the greatest- common-divisor matrix (gcd) defined on a gcd-closed set and on an arbitrary set of distinct positive integers. In fact, the evaluations of \textit{S. Beslin} and \textit{S. Ligh} [Bull. Aust. Math. Soc. 40, No. 3, 413- 415 (1989; Zbl 0675.10002)] and those of \textit{Z. Li} [Linear Algebra Appl. 134, 137-143 (1990; Zbl 0703.15012)] for higher-dimensional determinants are generalized, and the gcd is replaced by a function value, too, while the argument being the gcd. Let \[ \det_ IA={1\over n!} \sum_{(j_{11},j_{12},\ldots, j_{1n})}\cdots \sum_{(j_{r1},j_{r2},\ldots,j_{rn})} \prod_{k\in I}\varepsilon (j_{k1},j_{k2},\ldots,j_{kn})\prod^ n_{\nu=1}a(j_{1\nu},j_{2\nu},\ldots,j_{r\nu}) \] where \((j_{11},\ldots,j_{1n}),\ldots,(j_{r1},\ldots,j_{rn})\) run through the permutations of the set \(\{1,2,\ldots,n\}\) and \(I\subseteq\{1,2,\ldots,r\}\), \(\varepsilon(j_ 1,j_ 2,\ldots,j_ n)=1\) or \(-1\) according as the permutation \((j_ 1,j_ 2,\ldots,j_ n)\) is even or odd. The Dirichlet convolution of arithmetic functions (i.e., complex-valued functions on the set of positive integers) \(f\) and \(g\) is by definition \((f*g)(m)=\sum_{d\mid m}f(d)g(m/d)\). The main results are: Let \(A\) denote the \(r\)-dimensional matrix of order \(n\) given by \(a(i_ 1,i_ 2,\ldots,i_ r)=f(x_{i_ 1},x_{i_ 2},\ldots,x_{i_ r})\) \(i_ 1,i_ 2,\ldots,i_ r=1,2,\ldots,n\) where \((x_{i_ 1},x_{i_ 2},\ldots,x_{i_ r})\) is the gcd of \(x_{i_ 1},\ldots,x_{i_ r}\). Let \(I\) be a nonempty subset of \(\{1,2,\ldots,r\}\) with \(m\) elements. Then \[ \det_ IA=\begin{cases}\prod^ n_{i=1}\sum_{{d\mid y_ i\atop{d\nmid y_ t\atop t<i}}}(f\cdot\mu)(d), &\text{if \(m\) is even,} \\ 0, & \text{if \(m\) is odd;}\end{cases}\tag{I} \] \[ \det_ IA=\begin{cases}\sum_{1\leq k_ 1<\cdots<k_ n\leq n+s}(f*\mu(x_{k_ 1})\cdots(f*\mu))(x_{k_ n}) &\text{ }\\ \qquad\qquad\times \det_ I(E_ r(k_ 1,\ldots,k_ n))\text{det}_{\{1,2\}} (E_ 2(k_ 1,\ldots,k_ n)), &\text{if \(m\) is even}\\ 0, &\text{if \(m\) is odd,}\end{cases}\tag{II} \] where \(E_ r(k_ 1,\ldots,k_ n)\) denotes the \(r\)-dimensional \(n\times\cdots\times n\) matrix \[ E_ r(k_ 1,\ldots,k_ n)=(e_ r(i_ 1,\ldots, i_{r- 1})), \] \(i_ 1,\ldots,i_ r=1,2,\ldots,n\), and \(e_ r(i_ 1,i_ 2,\ldots,i_ r)=1\) if \(x_{i_ r}\mid x_{i_ 1},\ldots,x_{i_{r- 1}}\), and =0 otherwise.